Image forming apparatuses start their system after checking whole bits of their memories so as to confirm that the state of memories is normal.
On the other hand, as formed images become more higher-resolution, more colorized, and memory price is decreasing; the capacity of a memory in an image forming apparatus becomes larger.
Accordingly, error-checking on memory at power-on may take several minutes, which makes users uncomfortable.
Meanwhile, due to the progress of both the manufacturing techniques and inspecting techniques for memories, reliability of memories has been improved. Most of the reasons of errors detected by error-checking on memory are that memories deteriorated when heated for a long time and that memory terminals did not contact enough somehow when memories were added or pulled out.
For error-checking on memory at power-on, check patterns are used of only, for example, ‘01010101’ and ‘10101010’ in order to shorten time. In addition ‘read’ and ‘write’ are executed only once at the same address and with the same pattern for reducing checking time.
However, a memory error occurs, for example, because of the movement of charge leaking between a memory cell and its surroundings. The amount of leaking charge is different, depending on check patterns. Whether a memory error is detected or not depends on check patterns.
For this reason, there is a case where it can not be determined whether image degradation occurs due to memory error or not when abnormality occurs or image degradation occurs due to a memory error in image data areas.
In addition, error-checking on memory has to be done in detail for enhancing memory error detection. In this case, it may take more than ten minutes.
In regard to the above, JP 04-000542A discloses that memories are divided into a plurality of blocks and that error-checking on memory is executed on a memory block in sequence at every periodical timer-interrupt.
However, normal processes may become slow if error-checking on memory is frequently executed in detail by generating timer-interrupts, which brings uncomfortable feeling to users.
JP 03-171349A discloses that error-checking on memory is executed for an idle time between data processing by an apparatus.
However, it is not described concretively that what kinds of idle times are used, and if the error-checking on memory is cancelled before its completion, the result thereof might be different in some cases. To keep away from the result, the error check has to be continued, which makes users uncomfortable.
JP 2004-178387A proposes that, for reducing user's uncomfortable feeling, error-checking on memory is not performed in principle when a system starts and that error-checking on memory is performed only if the total processing number of sheets is more than a set number or if the capacity of memories is different from that at the previous power-on.
According to this method, it is possible to reduce user's uncomfortable feeling by omitting unnecessary error-checking on memory because error-checking on memory is performed only at higher incidence of memory failure.
However, the incidence of memory error depends on manufacturers or manufacturing date of memory devices, and early failure of memory device depends on manufacturing conditions.
In a case where image quality degrades because of memory error in an image data area and in a case where a processor has gone haywire, it can not be determined for a long time whether the cause is a memory error or not, and image degradation is so partial that the image degradation can not be recognized for a long term.
Moreover, there is a case where only one bit failure of the whole memory area leads a critical situation. The larger the capacity of memories is, the more the occurrence of a memory error increases in error-checking.
In addition, even if the period from turning on to turning off the power of an image forming apparatus is the same, the number of used sheets of paper is different, depending on users or dates, so that a margin of safety has to be made larger by lowering the setting of the total number of used sheets. If the total number of sheets exceeds the set value, error-checking on memory is executed at every power-on.
Furthermore, if a processor has gone haywire, a heater in a fixing device and a sheet feeding motor may keep power-on. Therefore it is desirable to take safety measures more certainly.